Build A Machine Learning Web App From Scratch

Build a Machine Learning web application from scratch in Python with Streamlit. We use real world data to build a machine learning model. In the first part you learn how we analyze the data and build our model, and in the second part we build the web app using streamlit.
Get my Free NumPy Handbook:
www.python-engineer.com/numpy...
✅ Write cleaner code with Sourcery, instant refactoring suggestions in VS Code & PyCharm: sourcery.ai/?... *
⭐ Join Our Discord : / discord
📓 ML Notebooks available on Patreon:
/ patrickloeber
If you enjoyed this video, please subscribe to the channel:
▶️ : / @patloeber
Code: github.com/patrickloeber/ml-a...
Resources:
Streamlit: streamlit.io/
Stack Overflow Survey Data:
insights.stackoverflow.com/su...
insights.stackoverflow.com/su...
Anaconda Tutorial: • Anaconda Tutorial - In...
~~~~~~~~~~~~~~~ CONNECT ~~~~~~~~~~~~~~~
🖥️ Website: www.python-engineer.com
🐦 Twitter - / patloeber
✉️ Newsletter - www.python-engineer.com/newsl...
📸 Instagram - / patloeber
🦾 Discord: / discord
▶️ Subscribe: / @patloeber
~~~~~~~~~~~~~~ SUPPORT ME ~~~~~~~~~~~~~~
🅿 Patreon - / patrickloeber
#Python
pandas scikit-learn numpy streamlit sklearn
Timeline:
00:00 - Introduction
00:39 - Project Overview
04:48 - Data Cleaning & Model
30:55 - Build The Web Application
----------------------------------------------------------------------------------------------------------
* This is an affiliate link. By clicking on it you will not have any additional costs, instead you will support me and my project. Thank you so much for the support! 🙏

[@patloeber]

If you follow this tutorial you have a great portfolio project for your resume: It shows working with real world data and putting it into a ML web app that visualizes the results. Hope you enjoy it :)

[@MrAntivirus66]

Hello Patrick, you are treally amazing.. i do have faculty project that i cant solve. that is face recognition using CNN's, and integrating it into django web app,, i am ''junior'' so it is really big step for me.. so if you could help with anything it would be great.. Ty kind sir :)

[@prajwalbagchi3965]

I am unable to solve an error in the 78th code box of your jupytor notebook which gives me the error:- first line:-"y contains previously unseen labels: 'United States" " 2nd line:- " y contains previously unseen labels: '2' " . I am unable to find a solution to this error, it would be of great help if anyone could guide me. THANKS😊

[@Saffron_SV]

@@prajwalbagchi3965 check the label must be changed to 'United States of America'... and for checking write this code df["Country'].unique()

[@ganeshchowdhary7479]

​​@patloeber hello sir I need to ask u some thing in the 10th cell what is the use of filtering the Employment column if ur gonna drop it in the next line it self 🤔 12:36

[@comfortrwuaan]

​@Saffron_SV Thank you for this, I was also stuck here for days

[@gabrielalopezgutierrez8871]

This is exactly what you are looking for if you are a junior data scientist who just started a web app project. Incredibly simple and yet complex intro.

[@Mike-jr7re]

What a maestro! Keep your great work, man. Really appreciate this high quality stuff!

[@yevg3907]

As I said previously I was probably within your first 1k subscribers and man I just love watching you go at these in the zone! Great content relevant no fluff just keep going love it

[@shishirrd]

This was fantastic! I built my first Python web app using exclusively this tutorial. Thank you, Python Engineer.

[@patloeber]

Great to hear!

[@yarlagaddaharish8137]

hey what should I include in .pkl file?

[@mdjisanahmed4310]

@@patloeber Is this annual salary?

[@ganeshchowdhary7479]

​@@patloeber hello sir I need to ask u some thing in the 10th cell what is the use of filtering the Employment column if ur gonna drop it in the next line it self 🤔 12:36

[@ganeshchowdhary7479]

​@@patloeber​​ hello sir I need to ask u some thing in the 10th cell what is the use of filtering the Employment column if ur gonna drop it in the next line it self 🤔 12:36

[@AhmadIHafez]

Thanks a lot it was detailed enough to help!

[@raymondmichael4987]

I'll come to this, your teaching made me think about starting to learn streamlit

[@prettiestthing]

This was amazing!! Thanks for this tutorial, loved it

[@moviesaddahindimovies2963]

No such file or directory: 'saved_steps.pkl' please help

[@tecnom7133]

Thanks man , waiting for more

[@tinylittleanj2]

This is amazing! Thanks for sharing

[@aikeshakeutayeva8828]

Thank you! That was interesting 💯

[@sulekhamohamud9085]

Thank you,this project actually works.Much appreciationn😇

[@peterodesola5056]

i really enjoy the tutorial, thanks sir.

[@NePhoosStAr]

Omg thanks you soo soo so much, because I was in college try Harding to understand the concept of machine learning. Got to learn so much. Like the concept of : data cleansing, data training fitting, how to use trained AI to make prediction. Your explanation is just mind blowing. I did this project and on top added my own touch which I would've never done if were to just to try to learn from College alone.

[@robertoespinoza199]

Amazing video bro!!!! Thanks so much for sharing your knowledge

[@patloeber]

thanks a lot :)

[@perceptron5983]

Cool. Added to my watchlist.

[@kevinalexis9886]

Thanks, this was very helpful!

[@chouchanewifak4351]

can you tell me why this is incorrect ?? X[:, 0] = le_country.transform(X[:,0]) X[:, 1] = le_education.transform(X[:,1]) X = X.astype(float) X

[@camilalissetsaenzvelasquez3981]

this is awesome! Thank you

[@gianlucavannoli2698]

such a good video. Thanks for sharing bro

[@kevinmeyer3863]

Great material! Would you mind making a future video on other data preprocessing/cleaning techniques? Would be great to systematize an approach.

[@patloeber]

yeah I want to do this in the future

[@sujithrasankar3438]

Awesome tutorial!!

[@gingerdragon6216]

What an awesome tutorial.

[@m-adem]

Thank You 🙌 You Are My Hero 💖

[@antonietakuz5636]

@Patrick Loeber Hola!!! :) greetings from Argentina, excellent video and explanation, please continue doing them

[@faisalsaddique3323]

THANK YOU SO MUCH SIR!

[@sadie-je4dl]

Awesome explanation.

[@lb9774]

Huge perfect and get new skill . Thank you.

[@patloeber]

glad you like it

[@saurrav3801]

It's nice to see you in this video bro......nice video too..bro..keep going 🔥🔥🔥

[@patloeber]

thanks :)

[@jaggyjut]

This is awesome. Could you create a tutorial on how to build a web app which can identify an image deployed on google cloud. One of the best ML channels. Pure Gold.

[@patloeber]

Thanks! Watch my latest tutorial, it's about Google Cloud deployment ;)

[@luckychitundu1070]

Great work

[@siddharthgawande1799]

Really awesome tutorial Thanks for this one, it helped a lot

[@shilpikumari4368]

Hello

[@shilpikumari4368]

have you made this project ??

[@siddharthgawande1799]

Yep

[@blogactivities5462]

thank you Patrick

[@DeepFrydTurd]

Awesome content 🙂👍

[@azrflourish9032]

WOOOOOOOOOOOWWWWWWWWWWWWWWWWWWWWWW!!!!!!!! YOU ARE GREAT! STREAMLIT ALSO AWESOME! IT MAKES EASIER TO LIVE :D

[@sriharishpopuri2821]

This is a helpful tutorial. Thank you. Can you also make a video on Interactions between different charts using streamlit? This will be useful to many(like me) I guess👍🏻

[@patloeber]

i'll have a look at it

[@jaxsamir3252]

You prolly dont care but if you are bored like me atm then you can watch all of the latest movies on InstaFlixxer. Have been watching with my gf during the lockdown :)

[@lucasxavier8072]

@Jax Samir yea, been using instaflixxer for since november myself =)

[@jackreacher5257]

Thank you

[@chineduezeofor2481]

Excellent video!!!

[@patloeber]

Glad you liked it!

[@Bianchi77]

Nice video, thanks :)

[@patloeber]

Glad you like it

[@kaushekravi2901]

Hi! Shouldn't the label encoder transformers need to be used only in the target variables? But in this video you are using it for the input variables too. Correct me if I'm wrong.

[@dhirajmohite5693]

Fantastic video

[@philtoa334]

Really good project.

[@patloeber]

Glad you like it

[@mikiallen7733]

Great tutorial, however in which step shall I add a streaming source datasets instead of the static one which you have presented here ? Best

[@ashanbandaranayeke9344]

Thanks Patrick! Was a lovely tutorial and guessing the frame can be used to design other linear regression apps easily (and other models with a little bit of work).

[@Sece1]

Great video! Can you explain from where you saved the pickle file and deployed the model? I am trying with 14 variables but I dont want to make prediction using all of them. How would that be possible?

[@Jimbob2k]

Great tut and great accent

[@CodeWithTomi]

Niceee... web dev with ML

[@patloeber]

Yep great combination :) Maybe one day I'll make a Django+ML project

[@CodeWithTomi]

@@patloeber that'll be great!

[@hardikrab]

This is amazing!!

[@patloeber]

thank you!

[@nishanttiwari9736]

Great Video

[@adityaariewijaya9284]

I like the way you set up the iterm2. Can you make a video how you set up that?

[@patloeber]

I pretty much followed this article: opensource.com/article/20/8/iterm2-zsh

[@briankoech4307]

Thank you very much. I learnt a lot following this tutorial. You are amazing.

[@patloeber]

glad it was helpful!

[@rangabharath4253]

awesome as always :-)

[@patloeber]

Thanks :)

[@Sidhubaba]

Awesome !

[@adeshmishra1671]

We want more videos like this..plzzz

[@patloeber]

sure :)

[@adeshmishra1671]

@@patloeber thank u sir🙏

[@dubglass]

Great material! What could be done to achieve a lower error ? That one obtained seems too big, don’t you think?

[@patloeber]

yeah it's not optimal, you can try different features, feature selection methods, normalization etc...and of course different models and optimize the hyperparameters

[@rololop34]

magnificent!

[@_rupeshchauhan]

Awesome Can you please make more videos on it. It's a request.

[@patloeber]

thanks for the feedback! I try :)

[@mohammedarmash8841]

@patloeber, what if do the label encoding in for loop ,what should i do after importing the model

[@ahmedhelal920]

Really thanks for this project , i have question please , on sidebar explore and predict the visualizations appeared on both of them . How can i solve that .

[@elad3958]

PERFECT!

[@rog0079]

You're a Gem !

[@patloeber]

Thanks :)

[@sushantgautam5535]

Great! Looking for more like this. Thanks.

[@patloeber]

Glad you liked it!

[@wilsongomes3360]

A truly masterpeace

[@taheralipatrawala7300]

How do we change the styling and structure of the web application as in HTML and CSS?

[@jayshworkhadka8406]

you used label encoder and gave that value, but what if we did one hot encoding, then how will we implement that in streamlit ?

[@karthikb.s.k.4486]

Nice can you also do interaction of FAST API with streamlit for this ML problem. Thank you

[@patloeber]

Yeah FastAPI is on my list

[@perceptron5983]

I second that. 😀

[@yevg3907]

Nice

[@crabosity]

Even after using terminal to install scikit learn jupyter does not detect it and says no module named sklearn?

[@adityasahu96]

Can you make a video on how to make front end using react instead of streamlit . Thanks

[@Dark-kj2fz]

Hellooo Sir, very nice tutorial. But how can we make it accessible to everyone (sort of url or website) not a localhost ? Thanks in advance and keep it up !!

[@divyansxii-b1677]

which type of dataset is this? supervised or non - supervised or semi supervised?

[@lightkira8281]

شكرا لك

[@techcompany_]

How we can calculate accuracy of above project??

[@fahadshaikh9099]

whenever i try to import streamlit as st it always shows error stating "import streamlit could not be resolvedpylancereportmissingimports" any idea what the issue can be? should i open vs code from the comman prompt that im using conda for ?

[@moeeztahir208]

Hello Sir, I had face a problem during practice regressor problem. I don't understand how to solve a problem ! (regressor is not defined)

[@frankkoslowski6917]

from sklearn import linear_model linear_reg = linear_model.LinearRegression()

[@adil288]

I am trying to deploy on streamlit cloud but throwing some weird error I guess I am using wrong versions in requirements.txt.Kindly let me know which version to be installed

[@tamirtsogbayar3912]

Thank you so much can you do fingerprint matching web ?

[@patloeber]

Will have a look at this

[@tamirtsogbayar3912]

@@patloeber thank you so much

[@placidaorlunwo5012]

I have three categorical data types and my confusion is on how to implement them given that this section of the code has just a two-dimensional array. ok = st.button("Calculate Salary") if ok: X = np.array([[country, education, expericence ]]) X[:, 0] = le_country.transform(X[:,0]) X[:, 1] = le_education.transform(X[:,1]) X = X.astype(float)

[@mavinii960]

This is related to the `LabelEncoder`: # Your DataFrame processing should look like this: le_education = LabelEncoder() df['EdLevel'] = le_education.fit_transform(df['EdLevel']) le_country = LabelEncoder() df['Country'] = le_country.fit_transform(df['Country']) # Prepare your feature matrix `X` and target vector `y` X = df.drop("Salary", axis=1) y = df["Salary"] Then you can go ahead and follow him at 21:42. Quick reminder that I am using a dataset from 2023, so in my Decision Tree Regressor, I needed to change the "United States of America" it is like: # Correct the country and education level based on how they were encoded new_data = [["United States of America", "Master’s degree", 15]] # Create a DataFrame for the new data point for easy manipulation new_data_df = pd.DataFrame(new_data, columns=["Country", "EdLevel", "YearsCodePro"]) # Apply the trained LabelEncoders to the new data new_data_df['Country'] = le_country.transform(new_data_df['Country']) new_data_df['EdLevel'] = le_education.transform(new_data_df['EdLevel']) # Ensure all data is in the correct numeric format new_data_df = new_data_df.astype(float) I hope it helps.

[@courageiroro5604]

Can we consider this prediction as Real Time prediction?

[@syaduinotech3681]

Thanks sir. Can you share the video how to published this kind of local data into github/ webapps that can be use by anyone in the world as apps/ web gui.

[@nyotowijaya7949]

You're the best

[@harshitpawar9038]

I can't understand one thing, how did you select those columns without any correlation?

[@necbranduc]

How well will this scale?

[@rengimcetingoz4740]

KING 👑

[@sireode]

When I recreated this app, when I predict based on Bachelor’s degree, the salary predicts well. But when I predict based on "Post grad " select, the salary prediction is not accurate. On zero experience, i get a a higher salary than when I have one year experience. This also applicable to the other selections. Only Bachelor’s degree select predicts well. Is this as a result of the outlier?

[@neotank874]

I ran into a problem. After creating the drop-down, whenever I select a list, the whole entire thing disappears.

[@openeyesvo8550]

can we build a ML project using jupyter Notebook? if yes how can i deploye on web app?

[@rahulpurashtu8904]

Why you used label encoder for country column ?. I think we have to use one hot encoder because it is nominal data.

[@erictovbin4200]

Ich kann nur danke sagen!

[@patloeber]

Gerne :)

[@pushkarprabhat3300]

Hey @Patrick Loeber can you provide the dataset of this project???

[@josephmargaryan]

How do we deploy the data so we can share it publicly with everybody

[@shubhampatrick]

I like this

[@shashikawiley6515]

What is the technique used here is it classification/clustering or any other technique

[@patloeber]

Regression

[@nikhilkshirsagar3995]

sir u haven't given the pkl file?

[@datainnovators1883]

where we collect the data set

[@shivagowrireddy139]

data = pickle.load(file) _pickle.UnpicklingError: invalid load key, '?'. can anyone help me out with this error?

[@sonyvishwakarma9821]

how to create a conda environment?

[@YASHPRASAD_]

Hey PE, I'm getting this error while installing the kernal in my virtual environment: "ipython: error: argument --user: expected one argument in anaconda" .................can you suggest/help how to solve this?

[@mohamedbarakat9908]

you should write " kernel " insted of " kernal " , it's a syntax error

[@ashishm2974]

How to use machine learning model (pickle / jooblib ) and make Graphs and log in /register and house price Prediction in django

[@hayki_ds]

is it possible to send the link someone? or it's working only local?

[@nishitbohra943]

can you tell me which machine learning algorithm is used in this code

[@patloeber]

LinearRegression and RandomForestRegressor

[@nishitbohra943]

@@patloeber thank you so much can u tell me specifically where each is used